diff --git a/llvm/lib/Transforms/Scalar/MemCpyOptimizer.cpp b/llvm/lib/Transforms/Scalar/MemCpyOptimizer.cpp
--- a/llvm/lib/Transforms/Scalar/MemCpyOptimizer.cpp
+++ b/llvm/lib/Transforms/Scalar/MemCpyOptimizer.cpp
@@ -1199,12 +1199,14 @@
 /// In other words, transform:
 /// \code
 ///   memset(dst, c, dst_size);
+///   ...
 ///   memcpy(dst, src, src_size);
 /// \endcode
 /// into:
 /// \code
-///   memcpy(dst, src, src_size);
+///   ...
 ///   memset(dst + src_size, c, dst_size <= src_size ? 0 : dst_size - src_size);
+///   memcpy(dst, src, src_size);
 /// \endcode
 bool MemCpyOptPass::processMemSetMemCpyDependence(MemCpyInst *MemCpy,
                                                   MemSetInst *MemSet) {
@@ -1250,8 +1252,21 @@
     if (auto *SrcSizeC = dyn_cast<ConstantInt>(SrcSize))
       Alignment = commonAlignment(DestAlign, SrcSizeC->getZExtValue());
 
+  // SrcSize must be available for the pointer/size calculations. Therefore we
+  // insert the new memset immediately before the MemCpy. An alternative would
+  // be to insert it just after the MemCpy.
+  // FIXME: If we want to insert the new memset above the memcpy, then we might
+  // wannt skip backward past any dbg instructions that are immediate
+  // predecessors to the memcpy. Any such dbg instructions were between the
+  // memset and the memcpy originally, so sinking the memset past them seems a
+  // bit reckless.
   IRBuilder<> Builder(MemCpy);
 
+  // FIXME: For debugability etc. it might be better to use the debug location
+  // of the old memset for the code emitted here rather than using the debug
+  // location from the MemCpy as we do right now.
+  //   Builder.SetCurrentDebugLocation(MemSet->getDebugLoc());
+
   // If the sizes have different types, zext the smaller one.
   if (DestSize->getType() != SrcSize->getType()) {
     if (DestSize->getType()->getIntegerBitWidth() >
@@ -1275,9 +1290,8 @@
 
   assert(isa<MemoryDef>(MSSAU->getMemorySSA()->getMemoryAccess(MemCpy)) &&
          "MemCpy must be a MemoryDef");
-  // The new memset is inserted after the memcpy, but it is known that its
-  // defining access is the memset about to be removed which immediately
-  // precedes the memcpy.
+  // The new memset is inserted before the memcpy, and it is known that the
+  // memcpy:s defining access is the memset about to be removed.
   auto *LastDef =
       cast<MemoryDef>(MSSAU->getMemorySSA()->getMemoryAccess(MemCpy));
   auto *NewAccess = MSSAU->createMemoryAccessBefore(